Register now After registration you will be able to apply for this opportunity online.
Trust but verify: Machine learning models for life cycle assessment of chemical processes
This semester project (also possible as a bachelor thesis) aims to validate our latest model for life cycle assessments of emerging chemical technologies.
Keywords: Process engineering, Life Cycle Assessment, Machine learning, Model validation, Chemical industry, Process Descriptions
**Background:**
Predicting the performance and environmental impacts of chemical processes requires a lot of information about the process that is usually not available at an early stage of research and development. At EPSE, we try to close this gap of missing data through machine learning driven estimation methods to predict process energy demands and raw material inputs. When working with machine learning models, it is key to validate your results on data unseen for training and hyperparameter fitting. This work aims to take a step beyond training, test and validation set splitting and investigates the performance of our machine learning-model when external data is used as an input. The external data is of the same general quality as our data set used to train and validate the model but is derived by another group of researchers who used different assumptions and background information. Thus, we can further investigate the applicability of our model to a broader range of input data.
**About you:**
- Student of mechanical engineering (e.g., energy flows & processes), chemical engineering, chemistry, or a comparable subject
- Good understanding of chemical processes and unit operations
- Coding experience, ideally with Python, would be a plus
- Independent and goal-oriented working style
**Background:** Predicting the performance and environmental impacts of chemical processes requires a lot of information about the process that is usually not available at an early stage of research and development. At EPSE, we try to close this gap of missing data through machine learning driven estimation methods to predict process energy demands and raw material inputs. When working with machine learning models, it is key to validate your results on data unseen for training and hyperparameter fitting. This work aims to take a step beyond training, test and validation set splitting and investigates the performance of our machine learning-model when external data is used as an input. The external data is of the same general quality as our data set used to train and validate the model but is derived by another group of researchers who used different assumptions and background information. Thus, we can further investigate the applicability of our model to a broader range of input data.
**About you:**
- Student of mechanical engineering (e.g., energy flows & processes), chemical engineering, chemistry, or a comparable subject
- Good understanding of chemical processes and unit operations
- Coding experience, ideally with Python, would be a plus
- Independent and goal-oriented working style
In this work, you will validate our latest machine learning-based method to estimate process energy demands on a recently published set of process simulation data. Furthermore, you will discuss limitations of our model and define a range in which the model should be applied.
In this work, you will validate our latest machine learning-based method to estimate process energy demands on a recently published set of process simulation data. Furthermore, you will discuss limitations of our model and define a range in which the model should be applied.
ETH Zurich
Tim Langhorst
Doctoral student
CLA F 15.1
Tannenstrasse 3
8092 Zurich, Switzerland
tlanghorst@ethz.ch
www.epse.ethz.ch
ETH Zurich Tim Langhorst Doctoral student
CLA F 15.1 Tannenstrasse 3 8092 Zurich, Switzerland
